A/B testing tools split website or product traffic between versions and report which one wins.
This guide ranks eleven products on what the first ninety days cost, the price once traffic and extra modules grow, how test data leaves, and whether you can buy and run it without a partner. Five are aimed at engineering teams more than marketers.
Vendors can pay for visibility on this page. It never changes what an entry
says about a product, including the criticism, and we earn nothing when you click through to a
vendor. How that works.
In short
What A/B testing tools software does
A/B testing software shows variants of a page or feature to separate visitor groups, measures a goal and uses statistics to judge a difference, usually priced by traffic.
In this order: setup effort, what it really costs, how your data comes back out, whether
you can leave, and who each A/B testing tool is built for. One rule sits on top of those tests:
Places 1 to 3 go to challengers, vendors outside the best-known brands in a category, wherever the category has them; the established names follow from place 4 in their ranked order. Why those five, and why there is no
score out of ten, is on the how we work page.
Quoted per organisation; plans packaged individually
—
Enterprise experimentation sold as individually packaged plans, no list prices
—
Country is where the vendor is headquartered or contracts from, which is a
different question from where your data is hosted. Where the two tell different stories, the
entry says so.
Website A/B testing with unlimited traffic on a published monthly price
Ranked #1 of 11 in Best A/B Testing Tools in 2026.
Published pricingEurope
Both paid plans keep traffic unlimited, which separates Varify from the tested-user tiers used by Convert and Mida. The first plan allows five experiments at once, supports A/B, split URL and multi-page tests and links to GA4.
Pro adds Bayesian statistics, full targeting and unlimited team members. The vendor lists cookieless tracking as an option, servers in the EU, and support within four working hours.
What stands out
Unlimited traffic
EU servers
Visual and code editor
Where it costs you
Entry plan covers a single domain and five active experiments, so an agency or multi-brand shop needs the higher plan.
AI features are billed in credits booked on top of the plan price.
Right for
Marketing teams testing one or two shop domains who want traffic-independent monthly pricing
Wrong for
Product teams needing deep server-side or warehouse-native analysis
GermanyPublished monthly plans; 15-day trial; AI credits booked separately
Warehouse-native experiments and flags you can also run on your own servers
Ranked #2 of 11 in Best A/B Testing Tools in 2026.
Free tierSelf-hostablePublished pricingElsewhere
GrowthBook reads results from your own data warehouse instead of collecting events itself, and its code is available under the MIT licence apart from separate enterprise directories.
The free plan caps users at three and projects at one, but experiments, flags and traffic are unlimited. Approval workflows, SSO and SCIM are Enterprise only. It suits teams with an engineer and a warehouse, less a marketer wanting a point-and-click editor alone.
What stands out
MIT-licensed core
Self-hostable
Warehouse-native
Not stated by the vendorFree plan for three users; Pro published per seat; Enterprise quoted
Lightweight website tests priced by tested users, with a sandbox plan
Ranked #3 of 11 in Best A/B Testing Tools in 2026.
Free tierPublished pricingAsia-Pacific
The sandbox plan has 100,000 monthly tested users and 30 monthly AI credits, which lets a small site run real tests before paying.
Growth removes the test cap, keeps data without limit and adds advanced targeting and integrations; it starts from a listed monthly price billed annually. Mida also offers native iOS, Android and Flutter SDKs and a Figma tie-in. Its privacy policy gives a Kuala Lumpur address.
What stands out
Sandbox plan
Tested-user pricing
Mobile SDKs
Where it costs you
The sandbox allows three live tests and keeps data for 60 days.
SSO and a service level agreement are reserved for Enterprise.
Right for
Small shops and agencies starting page tests on a low or zero budget
Wrong for
Buyers who require an EU-based contracting company only
Tested-user pricing from the first plan, with servers in Frankfurt
Ranked #4 of 11 in Best A/B Testing Tools in 2026.
Published pricingNorth America
Convert publishes prices by monthly tested users and offers a 15-day trial without a card. Its entry plan covers A/B and split-URL tests, but multivariate, multi-page testing and feature flags begin one plan higher, so a team needing those pays the second price.
Data is stored on servers in Frankfurt. The seller is Convert Insights, Inc., a US company, and free migration from VWO appears in each plan's feature list.
What stands out
Frankfurt servers
15-day trial
Tested-user tiers
United StatesPublished plans by monthly tested users; 15-day trial; Enterprise on request
Experiment platform for product teams, built around group sequential testing
Ranked #5 of 11 in Best A/B Testing Tools in 2026.
Pricing on requestEurope
ABsmartly describes itself as a hub for hypotheses, results and decisions, with an activity feed and in-platform documentation. Its group sequential engine is meant to shorten tests without breaking statistical rules.
SDKs cover native apps, websites, email and algorithms, and a visual editor exists for no-code launches. The FAQ says test data can stay in the customer's own cloud account, and the company is Dutch.
What stands out
Group sequential stats
SDKs for many stacks
Annual contract only
Where it costs you
No price is published and only yearly subscriptions exist.
The only trial is a paid 30-day proof of value.
Right for
Product organisations running many parallel experiments across web, apps and email
Wrong for
Small sites wanting a monthly plan or a free trial
NetherlandsYearly subscription, price on request; paid 30-day proof of value
Experiments on top of feature flags inside a product analytics suite
Ranked #6 of 11 in Best A/B Testing Tools in 2026.
Published pricingNorth America
PostHog runs experiments as feature flags, so the first million flag requests each month fall inside the monthly allowance and experiments are billed with them. It offers Bayesian and frequentist engines and links session replays to each variant.
The code outside its ee directory is MIT-licensed. The drawback is that testing sits beside analytics, replay and surveys, which suits product teams more than a marketer who only wants page tests.
What stands out
Experiments on flags
MIT-licensed core
Bayesian and frequentist
United StatesMonthly usage allowance, then usage-based; published
French platform that now leads with an AI experiment builder
Ranked #7 of 11 in Best A/B Testing Tools in 2026.
Self-hostablePublished pricingEurope
Kameleoon's plans page prices its PBX Starter at 495 dollars a month for ten experiments and 50,000 visitors, then moves feature experimentation, personalization and mobile testing into an add-on bundle without a listed price.
Its trial lasts 30 days but allows three experiments, and data is not kept afterwards. Enterprise adds SSO, custom data residency and on-premise set-up. The company is French and its legal notice names Hetzner in Germany.
Product experimentation and flags, now part of the Amplitude family
Ranked #8 of 11 in Best A/B Testing Tools in 2026.
Free tierPublished pricingNorth America
The developer plan gives two million events a month, unlimited flag checks, 50,000 session replays and unlimited seats. Pro starts at a listed monthly fee with five million events and then charges per thousand events.
Warehouse-native deployment, SSO and HIPAA eligibility with a BAA are Enterprise features. The site states that Statsig is part of the Amplitude family, and the terms name Amplitude, Inc.
What stands out
Part of Amplitude
Warehouse-native option
Event-based pricing
Where it costs you
Pro bills per thousand events above the included volume, which makes cost hard to predict on busy sites.
SSO and role-based access control sit on Enterprise.
Right for
Product teams wanting flags, experiments and analytics from one usage-priced vendor
Wrong for
Buyers who need a vendor independent of a larger analytics group
United StatesFree developer plan; Pro published monthly plus event usage; Enterprise custom
Flag platform whose experiments are metered by monthly active users
Ranked #9 of 11 in Best A/B Testing Tools in 2026.
Free tierPublished pricingNorth America
The developer plan is free, has unlimited seats and includes 100,000 experimentation users a month. The Foundation plan starts with a 14-day trial, then bills ten dollars per service connection a month and about eight dollars per thousand client-side users, billed yearly.
Thirty SDKs cover client, server and edge. Enterprise adds advanced targeting and custom pricing. Session replay and AI run tracking sit in the same product.
What stands out
Usage-based billing
30 SDKs
Experimentation add-on
Where it costs you
Billing mixes service connections and client-side users, so a forecast needs both numbers.
Advanced targeting and some governance are Enterprise features.
Right for
Engineering teams that already release behind flags and want experiments on the same data
Wrong for
Marketing teams who want no-code landing page tests
United StatesFree developer plan; paid plan usage-based and published; Enterprise custom
Tiered web testing from Wingify, now united with AB Tasty
Ranked #10 of 11 in Best A/B Testing Tools in 2026.
Self-hostablePricing on requestAsia-Pacific
VWO sits under Wingify, whose site says the company has united with AB Tasty, and abtasty.com now redirects there. The pricing page names Growth, Pro and Enterprise but shows no prices.
Multivariate tests, sample ratio mismatch detection and guardrail metrics start on Pro, while SSO, API access and self-hosting are Enterprise only. Expect a sales call before you see a figure.
What stands out
Three plan tiers
Visual and code editor
Self-hosting on Enterprise
IndiaQuoted per organisation; Growth, Pro and Enterprise plans
Enterprise experimentation sold as individually packaged plans, no list prices
Ranked #11 of 11 in Best A/B Testing Tools in 2026.
Pricing on requestNorth America
Optimizely's pricing page says every plan is individually packaged and gives no numbers. Its experimentation product covers web, multi-page, feature and server-side tests with its own Stats Engine and warehouse-native analytics.
Contracts differ by region: EU customers contract with Optimizely AB in Stockholm, US customers with a New York company. For a small team the missing price list and the bundle with other modules are the main hurdles.
What stands out
Server-side testing
Warehouse-native analytics
Packaged per customer
United StatesQuoted per organisation; plans packaged individually
A/B testing software shows variants of a page or feature to separate visitor groups, measures a goal and uses statistics to judge a difference, usually priced by traffic. The differences that matter are rarely in the feature list, so this is
the order we would work through them.
01
Decide whether you need a published price
8 of the 11 tools here publish what they cost; the other 3 quote per organisation. The ones you can compare without a sales call: Varify, GrowthBook, Mida, Convert.com, PostHog, Kameleoon, Statsig, LaunchDarkly.
02
Decide how much the jurisdiction matters
These 11 vendors are established in 7 countries across 4 regions (North America 5, Europe 3, Asia-Pacific 2, Elsewhere 1). That decides whose disclosure law applies to what the vendor holds, wherever the servers are.
How A/B testing tools charge for traffic
Three meters dominate this market and they behave differently as a site grows. Convert and Mida count monthly tested users, Statsig counts events, and LaunchDarkly counts service connections plus client-side users. Varify is the exception: its plans state unlimited traffic and charge by plan tier instead.
GrowthBook charges per seat on its cloud plan, and PostHog bills flag requests after a monthly allowance. Optimizely and VWO publish no figures at all, so the cost only appears after a sales call. A fair comparison takes your real monthly visitor count and your number of simultaneous tests, then asks each vendor for the price at both numbers, including the plan that holds the features you need.
Write down monthly visitors who enter a test, not total site visits.
Count tests you want live at the same time and compare with plan limits.
Ask what happens to the price in a month with double the traffic.
What the entry plan leaves out
Entry plans rarely hold everything a testing programme needs. Convert puts multivariate and multi-page tests and feature flags on its second plan, and Varify keeps Bayesian statistics, full targeting and unlimited team members for Pro. VWO places multivariate tests and sample ratio mismatch detection on Pro, with API access and SSO on Enterprise.
GrowthBook keeps approvals and SSO for Enterprise, and Kameleoon moves feature experimentation and personalization into an add-on bundle. Statistics are the quiet differentiator: sequential testing, CUPED and sample ratio checks decide whether a team can trust a result after a week. Check the plan page for these words before you compare headline prices, because a cheap tier without them forces an upgrade within months.
Find which plan includes sample ratio mismatch detection and sequential testing.
Check whether multivariate and multi-page tests sit on your plan.
List who needs SSO or API access and which plan that forces.
Marketing-led or engineering-led testing
The products split into two camps. Varify, Mida, Convert and Kameleoon centre on a snippet and a visual or code editor, so a marketer can launch a page test. GrowthBook, PostHog, Statsig and LaunchDarkly start from feature flags and code, so tests are usually built by developers and read in a warehouse or analytics tool.
ABsmartly and Optimizely span both, with SDKs and visual editors. Buying from the wrong camp is the usual mistake: a marketing team handed a flag platform waits on engineers, and an engineering team handed a snippet tool cannot test server-side logic. Decide who writes variations before shortlisting, then trial that person's workflow, not the dashboard.
Name the person who will build each variation and test their workflow in a trial.
Check for server-side SDKs in your language if developers will build tests.
Confirm the editor can handle a single-page application if your site is one.
Trials, contracts and leaving
Trial terms differ more than prices do. Convert and Varify offer 15-day trials, Kameleoon gives 30 days but limits it to three experiments and does not keep the data afterwards, and ABsmartly sells only a paid 30-day proof of value with yearly contracts. GrowthBook and PostHog let you start on a free allowance, and self-hosting GrowthBook keeps data in your own infrastructure.
Leaving is rarely documented: ask whether experiment history and results can be exported, and whether the snippet can be removed without breaking pages. Contracting entity matters when your legal team reviews terms, since Optimizely issues terms from different companies by region and Wingify, the owner of VWO, now says it has united with AB Tasty.
Ask for export of experiment history and raw results in writing.
Check the trial length, its experiment cap and whether data survives it.
Read the terms for the contracting entity and notice period for cancellation.
What goes wrong most often when buying A/B testing tools software
Choosing on the entry price alone when multivariate tests, targeting or SSO sit two plans higher.
Starting a test on too little traffic, then calling a winner before sample ratio and run time checks pass.
Giving a flag platform such as LaunchDarkly to a marketing team that expects a visual editor.
Assuming a renamed or merged product keeps its old plans, since AB Tasty now redirects to Wingify.
07
Frequently asked questions
7 answers
What is the best A/B testing tools in 2026?
Varify leads our ranking of 11. Varify sells website tests on a flat plan with no traffic cap, so a busy shop does not pay more as visits grow.
The entry plan covers one domain and five concurrent experiments, while Bayesian statistics, unlimited team seats and API access start on the next plan up. AI features are billed as a separate credit package. Servers sit in the EU and the vendor is a German GmbH.
Which A/B testing tools publish their pricing?
8 of the 11, with the pricing model each one publishes:
Varify: Published monthly plans; 15-day trial; AI credits booked separately.
GrowthBook: Free plan for three users; Pro published per seat; Enterprise quoted.
Statsig: Free developer plan; Pro published monthly plus event usage; Enterprise custom.
LaunchDarkly: Free developer plan; paid plan usage-based and published; Enterprise custom.
The other 3 quote per organisation.
Is there a free A/B testing tool?
GrowthBook, Mida, Statsig, LaunchDarkly offer a free tier or a free self-hosted edition.
Where are these A/B testing tool vendors established?
In 7 countries across 4 regions: North America 5, Europe 3, Asia-Pacific 2, Elsewhere 1.
Varify: Germany.
GrowthBook: Not stated by the vendor.
Mida: Malaysia.
Convert.com: United States.
ABsmartly: Netherlands.
PostHog: United States.
Kameleoon: France.
Statsig: United States.
LaunchDarkly: United States.
VWO (now part of Wingify): India.
Optimizely: United States.
Which A/B testing tools can you host yourself?
GrowthBook, Kameleoon, VWO (now part of Wingify). The other 8 are hosted by the vendor only.
What should you use instead of Varify?
GrowthBook and Mida are the next two on this page. GrowthBook is for Warehouse-native experiments and flags you can also run on your own servers; Mida is for small shops and agencies starting page tests on a low or zero budget.
Who should not buy Varify?
Product teams needing deep server-side or warehouse-native analysis. Entry plan covers a single domain and five active experiments, so an agency or multi-brand shop needs the higher plan..
If your A/B testing tools product belongs among these 11, tell us what it does and who it is for. Inclusion is an editorial call; what a listing is and is not is set out under software advice.
